Base de Capital

Abstract

This study examined the variable error (VE) for the golf impact position of high and low skilled golfers. A consistent shot type was employed by 20 golfers from 2 skill groups over 10-15 shots with a mid-iron. Analysis reported similar positional variability across categories. Significant differences were found for the whole group in VE of the distance of various body segments to the ball (p<.001) where variability reduced across the distance of the ball to the midpoint of the stance, compared to the pelvis and shoulders. Alignment variability significantly decreased towards distal segments of the kinematic sequence with VE for the alignment relationships of pelvis-stance compared to shoulder-stance and shoulder-pelvis showing reduced levels (p<.001) in the low skilled golfers. Tilt variability in the frontal plane around the sagittal / anterior-posterior (AP) axis also presented significantly reduced levels from the pelvis to the shoulders (p<.001) with no effect of skill level. Conclusions suggest that coaches should pay particular attention to the variability presented at the distal end of the kinematic sequence and the alignment of the pelvis in relation to the stance.

Introduction

Most sports possess critical features that contribute to the successful performance of a skill (McPherson, 1990) and if modified too greatly these critical features can affect the outcome and can result in detriments to performance (Arend & Higgins, 1976). In ball sports these features are often most prevalent at the moment of impact between either a sports implement or human performer and the ball. Critical impact factors at the instant of contact govern the parameters of ball launch and flight and subsequent shot outcome.

In golf, the impact factors of angle of attack, centredness of strike (impact location on the clubhead relative to the centre of gravity), clubface alignment (in relation to the target line), path of the clubhead and clubhead velocity (Tuxen, 2009) present the areas in which coaches strive for reduced detrimental variability (Chapter 1.2; Langdown et al., 2012). Betzler, et al. (2012) have shown that more highly skilled golfers exhibited significantly reduced shot variability in clubhead velocity, efficiency of strike (resultant ball speed as a direct result of clubhead speed: coefficient of restitution), centeredness of strike, angle of attack, club path, and face angle at impact compared to less skilled golfers.

Reduced variability in factors governing movement outcome accuracy has also been found in table tennis (Bootsma & van Wieringen, 1990) and basketball (Miller, 2002). Miller (2002) reported a decreasing level of movement variability moving from toe to ball in basketball free throw kinematics, suggesting that towards the accuracy dominated end of the kinematic sequence expert performers present reduced variability. The impact position of a golfer’s body could follow this pattern

the accuracy of strike of the golf ball i.e. reduced variability from foot position through to the hand and clubhead positions (Chapter 1.2; Langdown et al., 2012).

Behaviour has been suggested to emerge from a system of constraints when performing a goal orientated task rather than a consistent motor programme (Newell, 1986). This would suggest that movement variability to meet a task outcome will exist. Such variability may be detrimental or functional to the movement outcome and it is important for golf coaches to understand the role that movement variability plays within the swing. Chapter 2 has shown that address variability in alignment between the shoulders and stance is reduced for more highly skilled golfers compared to less skilled golfers (Langdown, Bridge & Li, 2013a). Chapter 2 also reported that distance, alignment and tilt variability decreased across proximal to distal body segments when addressing the golf ball. These findings emphasise the importance of reducing variability in the body segments that have an increasing effect on performance accuracy.

Coaches (e.g. Hebron, 2001; Mann et al., 1998) suggest that the address positions a player adopts can influence impact with the ball. Mann et al. (1998) have suggested that the position of the ball relative to the stance has a critical influence over the angle of attack and that the body reacts in the attempt to “find” the ball with the clubhead, in fact they state that the ball position affects the swing more than any other alignment factor. As well as the angle of attack as suggested by Mann et al. (1998), there will also be other effects on factors such as club path and club face orientation at impact which will alter the resultant ball flight. Mann et al. (1998) continued by suggesting that compensations are important if the ball has been positioned poorly at address. This shows the importance of address and how

variability here will have subsequent effects on the position of the body at impact as the golfer solves the movement problem of returning the club to the ball to attempt to meet the task goal. The problem of address variability and consequential impact position variability becomes increasingly difficult when moving away from the flat surfaces of a range bay or the tee boxes on the course; research is yet to establish how golfers cope with this environmental factor.

Limited research exists into the variability of body positions at the moment of clubhead impact with the golf ball. Horan et al. (2011) reported that both male and female golfers showed decreasing hand position variability from the top of the backswing through to ball contact even though they used different coordination strategies to zero in on impact. Clubhead trajectory was also shown to follow a similar funnelling trend. Bradshaw et al. (2009) recently studied the kinematic variability that occurs within the swings of high vs. low handicapped golfers using 2D video analysis and found no differences between skill levels for the variability presented in the trunk, lead wrist and elbow angles at impact. However, it must be remembered that a true representation of the movement of the swing is only achievable through 3D kinematic analysis.

With suggestions that the impact factors form the critical features of golf swing performance (Chapter 1.2; Langdown et al., 2012) it is crucial that the movements of the body are also analysed to see if these are affected by skill level. The five impact factors are all influenced by the position the golfer sets at address to the golf ball and movements the golfer makes during the downswing which emerge from the environmental, organism and task constraints. However, it is

adjustments to the swing this has to be done within the backswing as the duration of the downswing to impact is too short. This notion is supported by research (e.g. Verbruggen & Logan, 2009) suggesting that typically, participants can inhibit or change a specific response when a stop signal is presented close to the moment of stimulus presentation (in this case the beginning of the swing), but they cannot inhibit or change their movement when the stop signal is presented close to the moment of response execution (i.e. the downswing moving towards impact) (Cochran & Stobbs, 1968). Other sports have shown that there is a “zeroing in” or “homing in” process that allows variability to be minimised towards the critical phase of performance (e.g. impact) (e.g. table tennis: Bootsma & van Wieringen, 1990; long jump approach phase: Lee, et al., 1982; Scott et al., 1997).

This study aims to inform coaches of how a golfer’s ability level may influence variability in impact positions and its changing magnitude in the proximal to distal segments of the golfer’s body at the point of impact between the clubhead and the golf ball. It is hypothesised that more highly skilled golfers will present reduced variability at impact compared to less skilled golfers across body position variables. If the critical features of the golf shot with a midiron are the impact factors then this would suggest that with increasing skill level will come increasing proficiency at presenting an impact position that maximises control of distance and accuracy towards a given goal. This hypothesis has also been founded on elite players in other sports exhibiting a funnelling or “zeroing in” of movement towards impact in the sports of table tennis (Bootsma & van Wieringen, 1990) and long jump approach run (Lee et al., 1982; Scott et al., 1997).

Method

Participants

A convenience sample of 20 golfers (17 male and 3 female) were recruited following local committee ethics approval and formed high and low handicap groups which included golfers between the ages of 18 years and 27 years (M ± SD: HSG = 21±1 yr and LSG = 21±2 yr). Participants were categorised by CONGU handicap (high (HSG): n=10 Category 1: handicaps up to 5.4 & professionals with a mean clubhead speed of 39.2 m/s. Low (LSG) n=10 Category 3: handicaps of 12.5-20.4 with a mean clubhead speed of 37.2 m/s). This method of categorisation is common practice within golf research but does not identify areas of specific skill that contributes to a lower handicap. It merely represents each individual’s overall skill levels relative to their home course. All participants reported that they were injury free and experienced no pain during their golf swing.

Procedure

All participants performed 10-15 swings with their own mid iron (6 or 7-iron) that were captured at 250Hz using a 13 camera VICON™ MX system and 34 reflective markers. All participants were asked to address the ball and perform their normal swing to produce a consistent ball flight with maximum accuracy and distance towards an identified target on a golf net. For a full description of the experimental procedure see Chapter 2; Langdown et al. (2013a).

Selection of Variables

Variables were selected based on address variables presented in Chapter 2 and Langdown et al. (2013a), where a panel of expert coaches from the Professional Golfers’ Association of Great Britain and Ireland membership were consulted on the importance of the consistency of a number of variables that measured a golfer’s position at address. It was important to see the influence of these positions at impact in order to assess the effect of various body segments in relation to the critical moment in the swing. The following variables (Table 3) were selected to be measured.

Swing Processing

After reconstruction and labelling of marker trajectories, each individual’s swing raw data files were exported from the VICON™ Nexus system as a comma separated values file. Functions and scripts were written for data processing in Excel and allowed impact to be identified through an average of 2 frames (1 prior and 1 post impact) or a single frame (where the clubhead was in contact with the ball) depending upon captured positions of the clubhead. Positional data analysis allowed all variables to be measured at impact over the series of shots played. Where markers were not visible at impact the trials were removed from analysis (2 trials).

Table 3 Static Variables Measured at Impact Position

Variable Definition

Distance of Stance to Ball True distance to the ball from the middle of the stance (determined by the midpoint of the vector between the 1st _{toe of each foot).}

Distance of Pelvis to Ball True distance of the centre of the pelvis to the ball (centre pelvis was taken as the midpoint of the vector between the left and right ASIS).

Distance of Shoulder to Ball True distance of the centre of the shoulder line to the ball (centre of the shoulder line was taken as the midpoint of the transverse plane projection of the vector between the left and right acromion processes).

Pelvic Tilt Hip tilt in the frontal plane around the sagittal/anterior-posterior (AP) axis (determined using the tangent of the transverse plane projection of the vector between the left and right ASIS).

Shoulder Tilt Shoulder tilt in the frontal plane around the sagittal/anterior-posterior (AP) axis (determined using the tangent of the transverse plane projection of the vector between the left and right acromion processes).

Pelvis-Stance Alignment Angular relationship of the pelvis and stance (open or closed). Shoulder-Stance Alignment Angular relationship of the shoulders and stance (open or closed). Shoulder-Pelvis Alignment Angular relationship of the shoulders and pelvis (open or closed).

Data Analysis

All data were checked for approximation to the normal distribution, and group and individual means and standard deviations were calculated. Natural log transformations were used to normalise for shoulder tilt, pelvis-stance alignment, and shoulder-pelvis alignment variable error, log to base 10 for pelvis and shoulder distance to ball variable error and a reciprocal transformation for shoulder-stance alignment variable error data. To assess the variability between trials (within golfer) and between levels of skill (HSG compared to LSG) variable error measures were used (Schmidt & Lee, 2005) where Variable Error (VE) = (√∑(xi – M)2/n) and xi is the value of the selected variable for each trial, M is the mean average of this variable for all shots played and n is the number of shots played. Mixed factorial MANOVAs (repeated factor: VE, group factor: skill level) were used to assess any differences in VE between variable groups of distance (a 3 x 2 MANOVA, impact variable x skill category using stance, pelvis and shoulder mid-points to ball), alignment (a 3 x 2 MANOVA, impact variable x skill category using angular relationships for pelvis-stance; shoulder-stance; shoulder-pelvis) and tilt (a 2 x 2 MANOVA, impact variable x skill category using pelvis; shoulder). When significant effects were observed post-hoc tests with Bonferroni correction were used. All data analyses were carried out with SPSS 20.0 and data are reported as mean and standard deviation unless otherwise stated.

Results

The mixed-factorial MANOVA analysis showed there to be a significant main effect across the whole group in VE (Table 4) of the distance (Wilk’s λ=0.33, F(2,17)=17.5, p<.001, ηρ²=0.67), alignment (Wilk’s λ=0.43, F(2,17)=11.21, p<.001, ηρ²=0.569) and tilt (Wilk’s λ=0.42, F(1,18)=24.5, p<.001, ηρ²=0.58) variable groups.

Distance VE

Overall whole group contrasts revealed no significant difference between any of the distance to ball VE variables (F(1,18)=.202, p=.659, ηρ²=0.01). A significant interaction was found between skill categories and distance VE (Wilk’s λ=0.611, F(2,17)=5.413, p=.015, ηρ²=0.389). The pattern of variability was different between groups, where the HSG showed a reduction of VE moving from the stance distance to the ball, to the pelvis and then shoulders. LSG presented a different pattern where the shoulders presented increased VE compared to the pelvis distance to the ball (Table 4). However, Bonferroni post-hoc comparisons presented no differences between groups in any of the separate distance parameter variable errors (stance: p=.351; pelvis: p=.273; shoulders: p=.295). Distance VE of the stance to ball exhibited the most variability in this group with the distance VE of both the pelvis to ball and shoulder to ball being approximately 68% of its value (Table 4). When VE was considered within each category of skill the same pattern was found: Stance distance VE was significantly greater than both pelvis (HSG = p=.035; LSG = p<.001) and shoulder distance (HSG = p=.015; LSG = p=.017) VE but there was no

significant difference between pelvis and shoulder distance VE (HSG = p=.227; LSG = p=.404).

Alignment VE

Overall whole group contrasts revealed no significant differences between the impact alignment VE variables (F(1,18)=.633, p=.436, ηρ²=0.03). A significant interaction was found between skill categories and alignment relationships VE (Wilk’s λ=0.57, F(2,17)=6.38, p=.009, ηρ²=0.43) showing that the pattern of variability was different. HSG produced a reduced shoulder-stance alignment relationship VE compared to the pelvis-stance and shoulder-pelvis, whereas LSG demonstrated the lowest VE in their shoulder-pelvis alignment relationship. Again, Bonferroni post-hoc comparisons presented no VE differences between skill categories in any of the separate alignment relationship variables (pelvis-stance alignment relationship: p=.069; shoulder-pelvis: p=.626; shoulder-stance: p=.274, Table 4). The largest alignment VE exhibited in this group was the pelvis-stance relationship, with both the alignment VE of the shoulder-pelvis and shoulder-stance being approximately 80% of its value (Table 4). When VE was considered within each category of skill, differences were found for LSG: a reduction in VE was found between shoulder-stance alignment compared to pelvis-stance alignment (p=.007, Table 4) and between shoulder-pelvis alignment compared to pelvis-stance alignment (p<.001). HSG presented no significant differences between all alignment variables.

Tilt VE

VE was significantly reduced in shoulder tilt compared to pelvic tilt at impact (Wilk’s λ=0.42, F(1,18)=24.5, p<.001, ηρ²=0.58). There was no interaction effect of group (Wilk’s λ=0.92, F(1,18)=1.57, p=.23, ηρ²=0.08) with both HSG and LSG showing similar VE levels in each segment.

Table 4 Mean ± SD for Variable Error (VE) for each Variable

Note. #£$ _{indicate significant differences between variables (p < 0.05).}

Impact Variable Whole Group VE High skill group VE Low Skill Group VE

Distance of Stance to Ball (mm) 11.01±3.05# _{10.23±2.48 11.80±3.48}

Distance of Pelvis to Ball (mm) 7.55±1.91# _{8.02±1.97 7.08±1.82}

Distance of Shoulder to Ball (mm) 7.52±2.31# _{6.97±1.79 8.06±2.73}

Pelvis Tilt (°) 1.23±0.49£ _{1.26±0.47 1.21±0.54}

Shoulder Tilt (°) 1.09±0.36£ _{0.96±0.22 1.22±0.44}

Pelvis-Stance Alignment (°) 2.31±0.93$ _{1.92±0.43 2.69±1.14}

Shoulder-Stance Alignment (°) 1.87±0.83$ _{1.67±0.66 2.08±0.97}

Discussion

This study has examined whether there were differences in the variability of specified body impact position parameters between high and low skilled golfers. Whilst overall VE in each group of variables was different between skills groups there were no differences between groups in the VE of the separate body position measurements at impact. Looking at Table 4 there was no consistent pattern of higher or lower variability between the groups and therefore the current work provides no support for the acceptance of the hypothesis that golfers of a higher skill level will show less variability. Results from Bradshaw et al. (2009) also suggest this, with no variability differences found between high and low skilled golfers at impact for trunk angle, and both lead wrist and elbow angles.

Horan et al. (2011) have demonstrated that although the male and female swings emerged with different upper body movement strategies during the downswing there was a common trend of the variability of hand and clubhead trajectory sequentially decreasing during the downswing to the point of impact with the ball. Other sports have shown similar patterns of funnelling towards impact with Bootsma and van Wieringen (1990) finding that the variation in the direction of a table tennis bat decreases toward the moment of impact with the ball in a forehand drive. This funnelling effect in golf has to be attributed to movements of segments of the body in relation to the task (i.e. to attempt to get the clubhead to strike the ball in a manner that achieves the goal of the task). The funnelling of movement variability of the golf club has to be initiated within the golfer’s body segments during the downswing. The results of the current study show that there is a significant

body, moving along the kinetic chain (Hume et al., 2005) of the golf swing. From the stance upwards, the segments that initiate the transition into the downswing (i.e. the pelvis (Cheetham et al., 2008)) present the largest amount of VE in position from trial-trial with the shoulders presenting the least.

The higher the skill of a golfer the lower the shot to shot variability in clubhead speed, efficiency of strike, centeredness of strike, angle of attack, club path, and face angle at impact (Betzler et al., 2012). The hands and clubhead positional VE were not analysed in this study but future research should aim to see if this trend continues across these segments and into the clubhead’s position and whether this is the critical feature that distinguishes between skill levels. Horan et al. (2011) have already established that these segments do not differ between male and females but only tested skilled golfers and so their results cannot be extended to other ability levels. If lesser skilled golfers are unable to continue the funnelling of variability to